inflated ratings

[a517dogg]

Inflated USFA ratings... what are your opinions. some people say there is no such thing, because the person earned their letter somewhere, etc etc.

just so yall know i am now the proud owner of an inflated rating.

[damianip]

Originally posted by a517dogg
Inflated USFA ratings... what are your opinions. some people say there is no such thing, because the person earned their letter somewhere, etc etc.

just so yall know i am now the proud owner of an inflated rating.

There are inflated rating, and there are deflated ratings. Some fencers are "Ds" when maybe, in another division, they might be a "C". Any system has its faults.

If we kept track of how often they re-earned their ratings, that might help a bit, but it's never going to be perfect because some fencers have trouble with certain fencers so a "C" is never going to consistently beat a "D" because the rating is vague.

Even if it were true, there are ranges within each rating. My coach, who just re-earned his "A" is an excellent fencer. However, I don't think he's going to be beating Cody Mattern or Soren Thompson (both "A") any time soon.


Paolo (I'm an inflated "E")

[Chris]

Originally posted by damianip
There are inflated ratings, and there are deflated ratings. Some fencers are "Ds" when maybe, in another division, they might be a "C".

This is literally true, as well as figuratively;
Since both the USFA, and Canadian systems allow people to be demoted, not based on their performance, but rather based on their (lack of) participation: this can easily lead to someone who used to be a 'B', or even an ex-olympian, fencing in Div 3, simply because they haven't been to any big tournaments for a number of years. Now, we probably wouldn't expect them to perform at the same level, BUT, let's just say that though I have been fencing locally, I got demoted to D02 last fall because there's only been one tournament a year that's higher that D1, and with several B's and 3 or 4 other C's, I've come up a little short...well, here we are in 2003, and someone may well have gotten their 'D' for the first time the other day, at the fabled Hangover Epee; technically, they willed be put higher, but who would you really expect to do better?

Any system has its faults.

A lot of the 'disparity' and incongruous results come from people being inconsistent, both from tournament to tournament, as well as from bout to bout, as well as differences in competition patterns:

Certainly, someone who fences three weapons/event in one day should maybe not expect to have their best stuff at the end of the day, even if they do this regularly; this can certainly be expected to lead to a 'surprising' result, when fencing a 'peer' who enters only one event, and having not spent themselves, comes to the piste a little fresher; likewise, someone who draws a tough opponent early, and succeeds, may well, also, be operating at a disadvantage when meeting their next opponent. (hence my displeasure with the new passivity rule, though, since I haven't had occaision to want to draw out a bout, my fears have not been realized, but if people are as consistent in interpreting this as they are in understanding the definition of attack in foil, I am certain it will crop up.)

If we kept track of how often they re-earned their ratings, that might help a bit, but it's never going to be perfect because some fencers have trouble with certain fencers so a "C" is never going to consistently beat a "D" because the rating is vague.

In Fact the Canadian system USED to do just that:
(NOTE THAT THEY ARE/WERE CONSIDERING CHANGING THEIR SYSTEM, IN SOME WAYS, MAKING CLOSER TO USFA: I HAVE NO IDEA OF THE STATUS OF THESE DEVELOPMENTS.)

they put a statutory limit on it: (nominally) 15% of the competitive corps is A's
15% " " " is B's
15% " " " is C's

There is expected to be some expansion of these groups over the competitive season, so at the end of the season, some portion (the 15% is allowed to 'stretch' a little, though I am not sure how much) of each group is demoted based on the fewest number of subject results in the past 2 seasons (I beleive it is 2).

Personally, I kind of like this, as unlike the 'Chess' style system, you aren't penalized for doing poorly, and you are definitley encouraged to compete as much as possible, since you want to post as many 'A', 'B', or 'C' results as possible to help secure your position: Under the USFA system (except for NAC/FIE points),unless you think an event can get you to the next level, the system really doesn't encourage you to fence, and since you only need 1 result per year to get said classification, the only benefit of 're-earning' it is the experience, development, and confidence (which is what is should really be about, anyway, but we all know better.), and (in essence) you really only create an opportunity for someone ELSE to pull them selves up a notch, and dilute the value of your classification.

Now most of us like fencing enough, and are good enough sports that we really don't think that way, or act on it, but the point is, the SYSTEM can allow that to manifest.

The OTHER thing I reallly like about the Canadian system is/was that there wasn't the catch-22 about knocking people out before the final. If you results indicate you've performed better than half of a certain letter-group would have been expected to, you retain that result: If you go to an event (I beleive there is a minimum size rule in both systems) with 1 A, 4 B's and 4 C's , USFA would require you to win for the B, and get 3rd (or tie) for a 'C'; Both systems say 'the results should look like this:
1. 'A' fencer
2. 'B' fencer
3. 'B' fencer
4. 'B' fencer
5. 'B' fencer
6. 'C' fencer
7. 'C' fencer
8. 'C' fencer
9. 'C' fencer
10. 'U' fencer

But it not likely that will happen; and Canadian system says in effect, if you do good, and beat them out in the standings, maybe (some of) you are better than we thought, and they 'chalk one up' for you. (note that this is NUMERICALLY done, and not exactly based on the results: Since, on this case, B's are expected to take 2nd, 3rd, 4th, 5th place, placing 2nd or 3rd, gets you the 'B', which, by the way, ALL the B's want to do, as well as the C's; though, clearly, the 'A' is just in it for the fun.
SO if it the top 10 looks like THIS:
1. 'B' fencer - 'chalks' the 'B' result
2. 'C' fencer - promoted to 'B'
3. 'B' fencer - 'chalks' the 'B' result
4. 'A' fencer - Drop kicks mask after loss to 'C'
5. 'B' fencer - Drop kicks mask after loss to 'C'
6. 'C' fencer - 'chalks' the 'C' result
7. 'U' fencer - promoted to 'C'
8. 'C' fencer - Schooled by 'A', and goes home.
9. 'C' fencer - Drop kicks mask after loss to 'U'
10. 'B' fencer - Drop kicks mask after loss to 'C'

Under USFA system, you might not even have to bother sending this result in.

I really like a lot about the (old) Canadian system, although, YES, it appears that it would lead to inflation, the annual demotion process takes care (some) of this ;(as I understand it) my only concern is that it would seem to lack 'granularity' (USFA system employs the year of the classification to achieve 25 levels of classification) by only using A, B, C.
I really think if they at least added a 'D' level, and also retained the year you earned classification, they could get a lot of granularity: if a someone got their current 'A' in 1991, and has maintained it ever since, why shouldn't that be reflected?

Additionally, USFA system doesn't really give you any credit for experience at all, and you can ride the benefit on one good result for years. We would certainly expect that, even if a 'C' were 'better' than some 'D', and might win 9 times of 10, there's always that 1 time the 'D' would win, and this might occaision itself on a date where it would matter; under the USFA system, this would help him for 4 years, under the Canadian system, it might only help him for the rest of the season.

[quote]Even if it were true, there are ranges within each rating. My coach, who just re-earned his "A" is an excellent fencer. However, I don't think he's going to be beating Cody Mattern or Soren Thompson (both "A") any time soon.[quote]

Another 'Canadian' solution might be to carry, along the the classification, the number of results you have acheived within the 2-year (or however long) review period: So if you have made 3 'B' results in this current, and previous seasons, combined, you would be class'ed 'B3', and hence enter the event above any 'B1', or 'B2', and below any 'B4', 'B5', etc...Thus increading a (supposedly desireable) higher degree of granularity to distinguish between the fencers.

Of course, if someone fences twice a year, and wins both to be an 'A3' they would be put below someone who fences 12 times a year, and might be 'A6'

No simple system can really be expected to adequately deal with significant differences in peoples' frequency of competition, and sphere of competition (i.e. I COULD go to the middle of North Dakota, and start teaching classes, and running tournaments. It wouldn't matter what kind of hacks we were, under the USFA system, if we get 64 warm bodies, and enough equipment, pay the dues, and fence enough tournaments, we can 'bootstrap' our way to a fist full of 'B' classifications: Once we go dive into the main stream, we'd probably get our asses kicked! [although, OTOH, no matter how GOOD we got, we'd never be able to go higher, certainly, at SOME point, you should have to step into the 'main stream' to see how deep it really is.])

IMHO, This is the one weak point of the old Canadian system, is that it requires interaction with classified fencers to get any classification, so you can STILL get into trouble where unclassified fencers can actually be pretty good, and 'through things out of whack'. Of course, that system allows for that to be remedied much more easily than the USFA system, and again, adding classifications down to 'E', and allowing an 'E' to be a 'bootstrap' classification could help, a lot.

(like anyone would listen to me...)
;-)

Paolo (I'm an inflated "E")

No, there are just a lot of E's who haven't been to the right tournaments, though it would seem you aren't one of them.

Have a nice day!

[swordsen]

Huh, I think I like the Canadian system better than I like ours.

[Sara]

well, im in the midwest, and its kind of hard for any woman to break a c d or b (at least in epee) b/c there arent enough women who show up to make it an a tournament. so it seems as if the only way to obtain a good rating is to either whip the men in a large open, or at a national tournament. oh well. either requires work. so yeah, we have a lot of "deflated" ratings i guess?(i'm a "d" who is tired of it!) :-)

[whtouche]

I think personally that the largest source of "inflated" ratings, at least locally for me, is when an open is mixed. The ratings chart doesnt look to the gender of the 'A's or 'B's present. As such, an uninspired mens turnout, mixed with women, can produce some guys with inflated ratings, people who wouldnt normally be able to get 'C's or the like wouldnt have as much trouble.
I hope this doesn't spark another men vs women gender equality debate

[a517dogg]

Chris i understood none of that.

yes, my inflated rating is the result of a mixed open. but i still beat two male Bs and a male A to win it, but i still feel its inflated. hmm.

[swordsen]

Hmmm You get on and basicly say that mixed events give inflated ratings becuase there are high rated women in the events. But you hope it doesn't start a gender war. sigh......
You must havea different sort of women in your division.

[oiuyt]

Not to weigh in on either side, but merely to add some facts, whtouche was probably referring to the sabre event held at the same time/place as a517dogg's epee event (and therefore saying nothing about your A).

The placements and genders of the sabre fencers were/are (new ratings follow old rating if a new rating was earned):

1 MALE B02 B03
2 MALE B02
3 MALE C02 C03
3 MALE C02 C03
5 MALE B02
6 MALE A02
7 MALE B02
8 MALE C02
9 MALE B00
10 MALE C02
11 MALE C01
12 FEMALE B02
13 MALE E02
14 FEMALE A02
15 FEMALE C02
16 MALE D02
17 MALE C99
18 MALE E02
19 MALE C99
20 FEMALE B02
21 FEMALE B02
22 FEMALE C02
23 MALE D02
24 MALE DO2
25 FEMALE D02
26 FEMALE C02
26 MALE U
28 MALE U
29 FEMALE U
29 FEMALE U

Yes, we just had a local sabre tournament with 30 people including 2 A's, 8 B's, 10 C's, 4 D's, 2 E's, and 4 U's. Yes we have a LOT of ratings in our division (as well as a lot of sabre fencers). That said, did mixing this event produce extra ratings? No. Only 3 ratings were earned, all of them merely updating of existing ratings. Did the women underperform men of similar ratings? Yes, for the most part. A couple of male C's finished below a female C (granted C99's vs. a C02). This was less true in the other weapons held the same day.

-B :)

[I see dead people]

chess rating system rules.

[oiuyt]

But requires a fair amount of overhead. I stopped playing competitive chess about 8 years ago, so some of the following might be slightly changed. At that time people did not know their exact ratings. Whenever you got a piece of mail from the USCF it included your current rating on the mailing list, and a quarterly book was published with everyone's rating. Now in theory they COULD post online, but do they? I doubt it. Additionally tournament organizers had to send in $0.50 for each game played in the tournament to pay for the data input for ratings. IIRC that was the price if sent electronically in the USCF's forms. If you used other forms but entered electronically it was some higher price, paper copies some even higher cost (reflecting the added costs to the USCF.

Given the average person in an event will fence, say 5-6 pool bouts and 2 DE bouts that would be an extra $3.50-$4 added to the cost of every tournament. Not all that much when you look at a NAC that already costs $40, but it's a significant addition to events that cost $10. Presumably this overhead is why the USFA HASN'T actually tried the chess system, at least on a trial basis (the BOD had, several (5ish?) years ago decided to experiment with it but nothing ever came of it).

So, yes, the chess system is good. The overhead is a MAJOR drawback.

-B :)

[a517dogg]

wtouche, inflated ratings can also come from a fencer who just has a good day. or a single upset, which happens alot in epee. opens dont necessarily have to be mixed to produce weird results.

[Catal]

Originally posted by a517dogg
wtouche, inflated ratings can also come from a fencer who just has a good day. or a single upset, which happens alot in epee. opens dont necessarily have to be mixed to produce weird results.

Or, a good fencer has a truly bad day. I got my "C," a year ago, on a good day- but man, oh, man did I have a bad time fencing last week. I feel pretty inflated now...

[whtouche]

Of course, there are many sources of 'inflated' ratings. I was only pointing out one that seemed most common in my area.

[edew]

Ratings (or classifications, as they're supposed to be called) are "inflated" if you put too much on them. All ratings are supposed to do is determine the seeding of the fencers, and the classification of the event. That's it. If you use the classification system as it's intended, then there's no "inflation". It's just is. If you want to make connotations of who should beat whom, the rating system is (and should be) woefully inadequate for that purpose.

There should be enough uncertainty in the ratings so that the results should not follow the expectations from the initial seeding so directly. Otherwise, why bother fence the bouts at all?

[epeemike81]

Originally posted by edew
Ratings (or classifications, as they're supposed to be called) are "inflated" if you put too much on them. All ratings are supposed to do is determine the seeding of the fencers, and the classification of the event. That's it. If you use the classification system as it's intended, then there's no "inflation". It's just is. If you want to make connotations of who should beat whom, the rating system is (and should be) woefully inadequate for that purpose.

There should be enough uncertainty in the ratings so that the results should not follow the expectations from the initial seeding so directly. Otherwise, why bother fence the bouts at all?

I think we should clarify the term "inflation". By the term inflation, I mean that there are many cases where it is easier to earn a rating now than it has been in the past. Earlier this year, I noticed that somebody had earned a b in my division who I didn't think should have. I then started noticing numerous C's, B's and A's being awarded to fencers who didn't seem on par. I am used to seeing some flukey ratings. that is to be expected with the system we use. however the level of "flukey" ratings has gone up so far that it is no longer possible to define them as flukes. that is to say that the average level of fencing required to earn an A, B, or C has decreased. Thus, you end up with more high ratings, which begets more high ratings. thus, inflation. Thus, it has nothing to do with who should beat whom, but rather with a qualitative analysis of all of the C's, B's, and A's in my division. Admittedly, this is all based on my own perception of this years results in NE and the surrounding divisions, but I believe many of my peers from those divisions would agree with me (feel free to chime in, guys).

-m

[Zelda]

After reading all of these, I think I like the Australian System. (despite its short falls.) We have a simple ranking system based on a rolling competition schedule. Someone explained it to me once, but I think it can be found somewhere on www.ausfencing.org either or, its seems to be a tad more fair then the american system.

[nahouw]

Originally posted by oiuyt
But requires a fair amount of overhead.
-B

When I play chess on freechess.org, it takes, oh, about 1/2 second to have my new rating calculated. Not much overhead at all.

For fencing, yes, for every bout, that would be alot of overhead; however, if you just use final placement, uploaded into a database with an algorithm, this could be done very quickly, and with a web-based interface would provide timely information across divisions, for those fencers who compete outside their division.

The problem is that many people like the current system -- there is too much critical mass to change the tide. I don't know why, because I had a kid that earned his C many weeks before in a tournament in another division. His card wasn't updated in the tournament he fenced with us. We sent the classification changes in with his old rating since we did not have proof of any change. At the next event, he brought proof from that division's website; now we have to send in a revised classification change report -- a time frame of many months. But this is the mechanism that many people prefer.....

[Inquartata]

Unfortunately, given that the USFA is run by a small group of overworked people, the chess system would be a nightmare. Consider how difficult it is for them to receive, log and confirm entries in a timely fashion; now you're going to expect them to run a classification system of this complexity?


Eric---Yes, but the ratings aren't only used for seeding purposes; for one thing they are also used to decide who gets to fence in which tournaments. So you could indeed have an ex-Olympian denied entry to a Div I NAC based on his/her failure to keep up a rating, no? Or permitted to enter a Div II/III NAC ( brr! ) on the same basis.

Even in seeding, they aren't really perfect for the purpose. You can get a great fencer seeded lower than is justified, which will distort the results in the pools...and, if the fencer in question behaves shrewdly in the pools, can distort the seeding for the DEs as well.

[edew]

Restrictions to NACs (which I think you are referring to) are not by ratings. One may have a lower rating and still be able to compete in Div I NACs. One only need some amount of national points (I believe). I can agree to the argument that the USFA itself is misusing the rating system. And I certainly agree that the pursuit of some mythical perfect rating system is both a worthless quixotic attempt, and really has little value even if found.

The coarseness of the existing ratings system is sufficient for the purposes we use them for: seeding events. At the DITD, I did poorly in my pools, being the #1 seed in my pool, and coming out third of six fencers (and placing myself 31st going into the DEs). I eventually finished third, which is actually a bit higher than I should be finishing, but then, that's the whole point of the coarseness of the seeding: there is uncertainty in the final results.

I personally don't agree with the HPC's use of classifications to restrict the size (and make-up_ of the Div I NAC competition. But, the choice they chose is sufficient as it does the job that it wanted: restrict the number of entries to a manageable size, while ensuring that those who do attend are all of a similarly high (enough) calibre.

I don't wish to go back to the discussion of chess rating systems versus fencing rating systems versus whatever else rat